Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
CK
1 0 5 0 5 2
MRC/(ABQ-R-1208
--_
,.
SIMULATION
AND THEORY
RADIAL EQUILIBRIUM
FINAL REPORT
OF
OF PLASMOID
PROPAGATION,
Mark M. Campbell
Randy M. Clark
Michael A. Mostrom
September
1989
41
_
Prepared
for:
Lawrence Livermore
Post Office Box 808
Livermore,
CA
National
Laboratory
94550
-=_
-
7"'
Under
Contract'
B052997
(1_
_- :i',_gg_
_.
_
Prepared
by'
MISSION RESEAR,Cft
cORPORATION
1720 Randolph
Road, SE.
Albuquerque,
NM 87106-4245
lm
E.,I,:_ l F_I_U_JC)N
Of= IF-IIC_
tDO_:Ut_'tL:2i,,IT
I_ I_jt_I_iMI_I-EID
iii
lli, rt'llrlili,
J{lii'rl4)
ll iln(lt, r Ilil'
li)
illl(tt, r t'iililrill'l
I lii_
(li;('lllllt,
iii
iiii_lilt'l'_
iii' lilt,
I,iil_rt,
lll't,
lilliliht,
r'ltli'-74(l._-l_;,_(i-4bl,
lll
J,itt, rlllllrt,
t_ii_i lirt.,lillrt,(I
ii_
I_._,
I)lolliirl
Niill(lllilJ
illi
ilt't'qilllll
(if
t_lrl_
Sl)(ill_ilirt,(I
li)Jill
ilt4t'lll')
llf lilt' I illill'(J
_l'llht.'r
lhr' I iililt'fl
_liilL's
(;lll'l, riiliil,lil
('lilit'lirlilii
t,llillhi)'l'i,s,
Illilkt's
,_,iili)
il'14111 Ihihllil)
Iliir
llll)
i,\llrt,_s
(lr inililit,
sihilil)
J'ilr ltir,
ilif(irlliilliiill,
iii
lhr'lr
l'iililllh!It'lll.',is,
ill)lilir.iil_ls,
ri'llrtsl,
lilS
lhill
ril41ils,
il('l_,, lirlli'i'_,s,
ils
llrllihlt'l,
use, _(lill(I
Iii, rr, hi
ilill
lqi fin)
llrllt'l,,_s
hifriill4t'
(lll_l'rlinil'lil
_liili,
tit Ilii,
ll(ht'rlislnl4
lit
(ii' ilillliilrs
I'iii_t,r_,ll)
lhr,
I!liht,r_Jl)
liri_iill'l)
t, Xlirt,_it,(I
iii' ('lilil'ilrnJii,
(lr lir(_(lli('(
t'lld(lrSi.'lll_'ll|
lllill
rt'iilJ
('lilllllrliJii,
(ili_t,
li('
'
iit,('t,_i rlinll,
ilSl,(I
/
i-
lht'
(iii lilll
lllll
iii Jlillll)
lhl, I !nill'll
lit, rr'hl
llllrli(ist,
liril(I -
tlliililil'ill'-
d ,_iliilt,_,
_lilill
lir
(il_llt'll
Irilili'llillrk,
(if
ill' Jill)
(IJ_i'hl_l,(I,
II
nl,
iii
iili,% ttlirl'lllll),
qir
I.iihilrli-
lil
J'iir
s,
li/
O_
ii
A
V
b
UCRL-CR--105052
DE91
000803
ABSTRACT
P
Cases _imulating
beam currents
tion and code speed-up were achieved for these cases. The net current
i
(I,,t ~ I_/_)
in the expected
continued
noise reduc-
scaling observed at
of the current
itself.
layer was
observed, however. A laminar flow model is developed that appears to give good agreement
with the simulation
results.
lr
i
t
MAS ER
4!;, 'i _,'",lBI....]-II('_N
(';[:
]ill
,..,
'.-;:
......
:..
ib/tlJE.t.:
::t.,l ] tS LJi',li..
CONTENTS
L
Section
Page
1,0
INTRODUCTION
2.0
3.0
4.0
CONCLUSIONS
31
5.0
FUTURE
32
MODEL
26
WORK
REFERENCES
33
Appendix
A
LOW-MODERATE
AT AFWL
BEAM-CURRENT
iii
SIMULATION
RESULTS
A-1
FIGURES
o
Figure
1
Page
Periodic simulation results comparing azimuthal magnetic field Be (B3,
0.6 kG) versus radial position X2 = R (X2, cre) at ct = 400 cm for two
runs with h = 1.8 MA, (a) run AA, (b) run AB (hollow beam), NOTE
CHANGE IN X2 SCALING.
Time averaged
h= 1.8 MA.
current
Time aver_.ged current profile for run AB (hollow beam) near end of
run with h = 1.8 MA.
Periodic simulation
for h = 7.2 MA.
cm
10
10
14
@
iv
11
12
13
Periodic simulation
for/b= 28.8 MA.
AD, axially-integrated
J, versus R at ct = 200 cm
16
Time averaged current profile for run AD near end of run with
I_ = Oo
.o.8 MA.
17
14
15
19
16
Simulationresuits
showingnet currentfordifferent
beam currents.
20
17
Simulation results showing the square root of the beam current for
different
beam currents.
21
Periodic
simulation
results
comparingtime historyand frequency
spectrum ofthe netcurrentfortwo runs.Ib= 1.8MA. Currentin
unitsof 1.35kA, frequencyinunitsofCOo= 3 x i0lrad/s.(a)run AA,
(b)run AB (hollow beam).
22
Periodic
simulation
AC, timehistoryand frequency
spectrum ofthenet
current.
I_= 7.2MA. Currentinunitsof1.35kA, frequencyinunitsof
tj0= 3 x 10Irad/s.
23
Periodic
simulation
AD, timehistoryand frequency
spectrumofthe net
current.
I_= 28.8MA. Currentinunitsof1.35kA, frequencyinunits
ofCOo= 3 x i0xrad/s.
24
Particle
distribution
pha,qe-space
plotofaxialmomentum (Pl - B,_/)
versusradialposition
(X2, cre).
Left-handplotissheathelectrons,
right-handplotisbeam electrons.
(a)and (b) 1.8MA run AB at
ct=400cm.
(c)and(d) 7.2MA run AC at ct=500cm.
(e)and
(f)28.8MA run AD atct= 200 cm.
25
Periodic
simulation
results
comparingB0 versusr atct= 3000 cm for
threeruns withI_= 22.5kA. (a)run AA, (b)run AB, (c)run AL.
A-7
18
19
20
21
A-I
18
T
E
A-2
Periodic
simulation
results
comparingE, versusr at ct= I000cm for
three, runs with I_ = 22.5 kA. (a) run AA, (b) run AB, (c) run AL.
A-8
A-9
A-4
Time averaged current profile in steady state for ,runs at I, = 22.5 kA.
A-10
A-5
A-11
OI
A-12
OJ
A-13
O_
A-14
O/
Periodic simulation results comparing E' versus r at late times for six
runs with h = 452 kA. (a) run BAl at ct = 2000 cre, (b) run BA2at
ct = 2000 cre, (c) run BC at ct = 2000 cre, (d) run BD at ct = 2000 cm,
(e) run BE at ct = 1600 cre, (f) run BF at ct = 1600 cm.
A-15
O/
A-3
A-6
A-7
A-8
A-9
A-10
,%-11
OI
Periodic
simulation
results
comparingaxially-integrated
J, versusr at
ct= 400 cm foreightrunswith I,= 452 kA. (a)run BA, (b)run BAl,
(c) run BA2, (d) run BC, (e) run BD, (e) run BE, (f)run BF, (g) run
= 452 kA.
Or
A-17
O_
vi
O
i
A-12
A-18
Time averaged current profile for run BB near end of run with
h = 1808 kA.
A-19
A-14
A-20
A'15
A-21
k-13
'
A-16
A-17
A-18
A'22
Periodic simulation BF, time history and frequency spectrum of the net
current, h = 452 kA. Current in units of 1.35 kA, frequency in units of
Wo = 3 101 rad/s.
A-23
Periodic simulation Bn, time history and frequency spectrum of the net
current. I_ = 1808 kA. Current in units of 1.35 kA, frequency in units
of w0 = 3 10l rad/s.
A-24
vii
b
1.0
INTRODUCTION
Using our earlier test data (reproduced in Appendix A), _rlorder to reduce noise in
the high-current cases we increased the number of particles per cell to 375 for the sheath,
used radial curreltt smoothing Over three adjacent cells and eliminated current inside the
axial cell (R _- 0) to reduce the fictitious spike near the axis. This decreased the fluctuation
amplitude by about a factor of three and made estimation of the average current values
from the plots much easier.
f
O
The summary plots shown here are in the same format as those in Appendix A but
continue the beam-current analyzed on to much higher levels. The trends are essentially
as expected.
e
We have completed a laminar-flow equilibrium model that predicts the radial profile
of the axial velocity of electrons and the relationship between azimuthal magnetic field
and radial electric field. Specific predictions are obtained for the net current and ion-beam
Q
temperature necessary for equilibrium. In ali cases the a_;:eement between the model
predictions and the ISiS simulations is excellent. The model also predicts parameter
scallngs which have not yet been tested in the simulations and, therefore, need further
verification.
2.0
Table
carried out in August 1989. We have also included a few older simulations
in some of the summary
plots.
In ali these runs, the axial velocity of the ion beam was
_,_ = ,0.62, the annulus between the beam and the conducting
was T_ = 43.2 kV
O
and one with the uninterzero net current,
replaced
with a particle reflecting metal boundary at R = 25 cm (see Figs. 1-5). These cases
agreed within ~ 10 percent, so higher currents were run with this feature which decreases
the simulations
115 MA case (which might require 15-20 CPU hours) was not undertaken
The width of the nonzero net current region, which had previously
-_ "h,_,_T"l/2,
continues
we were able
but a
and 13). The radial current profiles of these and some lower current
are superimposed
to run the 7.2 MA and 28.8 MA cases readily (Figs. 6-9 and 10-13, respectively)
as approximately
plasma at five times the beam density, and the ion beam temperature
(the same as concluded
to exhibit approximately
A negative
Noise
at higher currents.
This
The
At its
radial position there is some overlap with original beam electrons which have
O:
and loose ali their energy (Figs. 2la, c, and e). In this "mixing" region the sheath electrons
lose energy and axial momentum
The complicated phase space of the beam electrons (Figs. 21b, d, and f)'as they leave
radially is due to the field reversal of Er and Be in the sheath region just inside the beam
edge. Er and B0 have very similar radial profiles and go to zero at the same radial position
(Figs. 1 and 2, 6 an,_.7, and 10 and 11).
D
No significant compression or expansion of the ion beam was observed in any of these
runs, so we postulate that the equilibrium ion-beam temperature is relatively independent
of current.
O
Figures 18-20 show that a steady state has been rea_ed at least in the net current.
The noise has been substantially reduced over the level observed in the earlier simulations
reported on in Appendix A.
@
3
O
OI
O_
TABLE 1. SIMULATION
,,
@_
LLNL Plasmotd
Run
lD _
Ib
AA
-1.8 MA
AB
1.8 MA
AC
7,2MA
AD 28.8MA
AL
BE
BB
Propagation
Cell
Size
(cm)
0,1
0.I
0.05
0,025
X2MIN
-0,05
25
25
25
X2MAX
53.25
51.8
50.25
50.25
Cells
533
268
505
1010
Net CurrentResultsatDifferent
RadilrsPositions
Run ID
Radius
AL
BE
BB
AA
AB
AC
AD
54
i0.96:3:{}9"
-35.14
52.8
-9.57
51
40.17
50.3
56.16
50.125
-23,38
50
67.84
93.97
49
99.77 191.58 372.52
48
10.99 41.05
95.87
86.1
47.75
67.15
47.7
97.81
47.5
75.34
76.3
36.53
44
35,!
ii.51
40
9.02 12.1
10.82 -20,8
30
-0.67
-1.78
6,01
28
4.06
-I.II
-1.73
16
0.6
I 0.96
I
!
-9.8
-0.02
0,15
0)
@i
Ot
Oi
01
OI
.0,03
4
O
I 0 --I
4 .36
O
2.69
_I,
02
_ ' ,
--0
-2 64
.5
--0. I
', ','
1:
I 3. 3
26. 6
X2 _
'
'
39. 9
, ,
53.2
(a)
_21.%9
1 .62
II
"" 0
C::13
36
-0.91
0
-2.
18
25.0
,51 . 7
58.4
X2
45.
51 .8
(b)
Figure 1 Periodic simulation results comparing azimuthal magnetic field B0 (B3, 0.6 kG)
versus radialposition X2 = R (X2, cre) at ct = 400 cm for two runs with
0
4.98
--
--2.28
--..3.62
25.0
.31 .7
I
.38.4
X2
o,
I
45.
51.8
O"
(b}
. 10
-1
d 1-- I NT 1
1. 7 o
1_
1 .04
,._.
Zo
38
-0.29
,,_lJ
-0.95
O. 0
1
1._3..3
I
26.
X2
I
6
t _1
53. 2
.39 . 9
(a)
*1.I(_ -I
I
JI--INTI
I
I.,
0.42
_t
--
51 .8
e
_
'
im
--0.76
--
--1 .35
25.0
I
,.31 .7
I
,38.4
X2
4.5.
(b)
Figure
O
Iv
NOTE
CHANGE
IN X2 SCALING.
OI
0
8
r
9
0
'1
q_,
q,
5 1.%2'1 .........
-5.
12
25.0
_I
1
3 1 3
'
37.6
X2
.f
4.5.9
50.2
Figure 6. Periodic simulation AC, B, (B3, 0.6 kG) versus R (cm) at ct = 500 cm for
A = 7;2 MA.
0_
10
A
W
-O
45
18
'
'
[_O. 81
@
--1
.44
--
--2 ,07
25.0
,,
, I
31.3
37
x'2
43.9
50.2
Figure7. Periodic
simulation
AC, E, (E2,0.51MeV/cm) versusR (X2,cre)atct =
500cm forIb= 7.2MA,
@
@
11
@
_'1.(_ 7 "1
J l-INTI
I .32
_Zo. 43
I
q,-,.,.
-2.
18
0
-3.94
25:0
31 .3
37.
X2
43.9
50.2
12
_a
II
lS
II
,1_--1
9.
1
90
OI
_0
O0
-4
91
82 ....
25.0
I
51.5
I
57.6
X2
l
45.9
50.2
Figure 10. Periodic simulation AD, B0 (B3, 0.6 kG) versus R (X2, cre) at ct - 200 cm
for Ib = 28,8 MA,
O
O
14
1 .02
i
--0.86
--4.61
--
-6.4.8
25.0
1
,31. ,3
I
,37.6
X2
I
4,.3.9
50.2
Figure11,Periodic
simulation
AD, Er (E2,0,51MeV/cm) versusR (X2,cm) at ct=
200cm forI,= 28,8MA,
O
15
1.34
--1 .56
25.0
.....
jI-INT1
I ..........
I
51 . 3
i
,.57.6
X2
oi
45.9
50.
@i
0_
01
16
Q,
17
L_
Z
0
.................
io
__
.
_
N
o
(va) .LN_-t_n3
'
18
0
19
O
(v'4) _LN-a_sno
20
O
21
O
PLASMA
SHEATH
RUN
AAi
'_L
_L-'
VB,,m,63, _B_'Oc_2"NP'8,EIO,
VTH(2)'=O30
........
_/J,ii___
o,
oo
lO
.o!
Sp
..
1o 8
"7
10
.IT'
"_
1o '
2
100,
,
,
O0 L
'
i
0,74
,/_
,
1,47
2,21
2 '-J4
,Q
FREO
(a I
Figure
TIME-
400.20000
I8. Periodic simulation results comparing time history and frequency spectrum
of the net current for two runs. I_ = 1.8 MA. Current in units of 1.35 kA,
frequency in units of _)o = 3 l01 rad/s. (a) run AA, (b) run AB (hollow
beam).
22
PLASMA
SHEATH
RUN AC,
189,
VBI.B3.
NB-O,2.NPL32,ElO,VTH(2)"0,30
21480
'
142,
_ 9s
47,
'o,
10
1o 10
9i
11
_25,'
oo
_005
,1
25o.
'
"
_.47
2FREQ
95
4 42
590
500,
29920
Figure 19. Periodic simulation AC, ttme history and frequency spectrum of the aet; current.
h = 7.2 MA. Current in units of 1,35 kA, frequency in units of
w0 = 3 101 rad/s.
O
23
5o,
TIME=
1
_7_.
@
PLASMA
SHEATH
RUN
AD,
VB =-,63,
NB;6Oc_2"NP'I
,3E12,VTH(2)=O.
30
263
351,
_175
.=
88'
0'0,
,q
50,
....
iO0.
150,
_"/LA="
200,
1o
!
_=
10 0 5
1 ,47
2,94 _,,
F'REQ
--
4,42
TIME,=
@I
@_
5.89
200.
04000
Figure 20. Periodic simulation AD, time history and frequency spectrum of the net current. h = 28.8 MA. Current in units of 1.35 kA, frequency in units of
OJ
_0 = 3 x 101_ad/s.
@
@
24
ti
L .....
X2
'
21.
I .......
>'2
{e)
Figure
(f)
Particle distribution
phase-space
plot of axial momentum
(Pl =_ fl,_) versus
radial position (X2, cm). Left-hand
plot is sheath electrons, right-hand
plot is
beam electrons.
(a) and (b) 1.8 MA run An at ct - 4oo cm. (c) and (d) 7.2 MA
run AC at ct = 500 cm. (e) aad (f) 28.8 MA run AD at ct = 200 cm,
25
3.0
MODEL
@1
We propose a laminar flow model (similar to a model that has been used for magnetic
insulation
plasmold.
By "lamlnar"
equilibrium
In the
e_
of radius rb wereborn
with a conducting
of the electron
( = 0)
then
factor
,y-i=
(1)
where e (> 0) and m are the electron charge and mass. The ton-beam density n; and the
axial velocity v,; are assumed to be free parameters
v_ = 0 for r > rb. Conservation
independent
gives
eL
P, - ",/mu, - eA, = 0
(2)
A_ = 0 at the = 0 surface.
e_
E, = .- d_
=
dr
mc' d'_
e dr
by
(3)
rd"_
r_r
(0
'
n,-
e2 rdf
eornc21d
r_r
(d'7)
+ni
(5)
26
e
The azimuthal
magnetic
from Eq, 2
by
'
d
md
B, = -_.rA_, = -----("/u,)
e dr
(6)
e
Combined with Er from Eq, 3 this gives the radial fore.balance
which gives Er _ obe In agreement
equation
results,
E_ - v, Bo = O,
Ampere's
law relates
1_
= .o_(_,_.,- .,v.) ,
r dr (_B,)
e
Substituting
--
rld[d
dr
(_)
_ (_..)]/Zoe'
+ _rr_ (_,;..,-.,,..)= o ,
Substituting
(s)
nonlinear differential
equation
radial
e
For simplicity
1, Defining
(z < O)
(10)
y = c1" + c2e-_o To keep y finite for large negative
(r _. rh), we require c_ = 0, Outside the beam, v.i = 0 and n_ may be different thun inside
the beam, Defining A2 - n_(r > rh)/ni(r
@
d_v,
dx---T = A_v,
(x > 0)
(11)
27
whichhasthegeneral
solution
v,= 0ae
A'+ _4e"_',We assume(r_- rb)Awb/a
:_ i (many
plasmasklndepthsbetweenbeam and wall),To keepu, finite
forlargepositive
x, we
require
as= O,Thus,thevelocity
Isdescribed
by
_,=
(12)
_4e
"_" ' (_
> o)
(z<O)
Becausethepotentlal
and itsderlvatlve,
theelectric
field,
must be continuous
atJthe
beam edgex = O,thisdefines
thetwo constants
and Zlves
VA=
v.
---e'
(x < o,r < r_)
i - I + ),
--I-l+Ae -_" (_ > o,r >rb)
(13)
v,(r= rb) i
v,_
- ---1+A
(I,_)
whichIssmallIfA :_i,whilev,---+
v,iforA -,O,
OJ
Several relationships fbllow from this ann,lysis. From Eq, 6, the magnetic field at the
beam edge Is
mv,_ A
O,
Be(r= ,'b)
= e ,,i+ ,x
_b
(I_)
which vanishes for A --+ 0 and approaches a constant for A _ 1, The net current In the
beam Is
I
A
1'.. (r = rb) - 2___rb
BO(r : rb) = _ I._/_.,1 + A (rbCZb/C)
where
(16)
O"
28
O
I_ h(r+)=I+/_
"
'
(_)
(ts)
The azlmuthalmagneticfi,_+Id
exertsa confining
pressure
B_(r+)/2_o
on theionbeam
@
'
(19)
Al(1 + A),
i
These theoretical
predictions
with
The square of the net current lint, divided by the beam current
I_, gs plotted versus I_ for the simulation results In Fig, 17, and the result ts indeed constant
at approximately
5 kA independent
of beam current,
(here and in Ref, 1), we used an Ion drift velocity of J_i = 0,62, For ali but the earliest
'
simulations
In Ref, 1, we use a plasma density between the wall and beam equal to five
'@
value of 5,03 kA
for the constant of proportionality between I_t(r_) and I,. Moreover, Eq, lg gives a
predicted Ion-beam temperature for equilibrium of T# - 46.9 kV. In the ISIS simulations
In Ref. 1, we varied _ _zld empirically found that 43 kV gave a good equilibrium for beam
@
currents
research reported
here, this
same value of 43 kV was used for the ion-beam temperature in ali of the simulations and no
noticeable contraction
@
are excellent, particularly when the wide range of beam currents I, = 22.,5 kA-28.8 MA Is
considered,
29
A
The radial profile of the electron axial velocity In Eq, 13 also can be compared with
simulation, In ali of the MRG ISIS simulations, A was large (usually A_ = 5), The
description of the electron velocity v. in gq, 13 Is consistent with, the sin'ml_tion results in
that v./v.# was small at the beam edge _ = 0 and Increased rapidly over a few beam skirt
depths e/Wb until v. _ v._ tn the beam Interior, The near equality of Er and _B0 In the
simulations also ISconsistent with the laminar flow model used here,
Several
scaling
predictions
remaintobe verified,
One trendapparent
fromEq, 13,
but notyetverified
by computersimulation,
isthatvuat thebeam edgeincreases
as A
decreases and vj(r = rb) --* v_ as A _ 0, From Eq, 15, the magnetic field at the beam
edgeIs proportional
toA forA _ I,and saturates
ata constant
forA _ I,The constant
ofproportionality
iabetweenl_ei(rh
)and Ib,
is linearly proportional to the ion beam axial velocity _r._which was never varied from (},62
in our series of simulations, There is also a strong dependence of I_ on A, For A_ 1,
Ia Is proportional to A_--that ts, to the ratio of exterior plasma density to beam density,
Combined with Eq, 18, this predicts l..(r_) r #_/_(r > rb), or to the number of e,_terior
plasma skin depths across the beam, For A :_ 1, as in the simulations, I.et(rb) cz r#x/_,
or to the number of beam skin depths across the beam, The Ion-beam temperature for
3O
O
4,0
CONCLUSIONS
Increasing
thebeam currentto 28,8MA does not appeaxtostrongly
affect
theequlllbrlumscalingdiscussed
inour earlier
reports,While thereissome indlcation
thatthe
currentsheaththickness
isnow not decreasing
as rapidlyat higherbeam currentas at
lowervalues,
thenet currentstill
scales
asIn.,~ /_/._m,
The beam temperaturerequired
forequilibrium
appearstoremainconstantat43 kV independentofbeam current,
To trytoshed llght
on theobservedscallngs
intheISIScode results,
we completeda
laminarflowmodel ofthe electrons,
Specific
predictions
were obtainedforradialprofiles
aswellasforthenet currentand the ion-beaxn
temperaturenecessaryforequilibrium,
In
allcases,
the predictions
agreedverywellwith the ISISsimulations,
and new parameter
scalings
were broughtout which shouldbe verified.
31
A
OI
5,0
FUTURE
WORK
In spite of having carried out extensive and complex simulations over the last
two years, some of these parameters have remained fixed. The laminar flow model pre-
@l
sented here indicates several interesting parameter scallngs that, if verified, would further
our confidence in this model, We could continue simulations on to higher currsnts, but we
believe that this scaling is adequately
28,8 MA.
For instance,
Ucore" boundary
to larger radius
(i.e.,
temperature
instability
sufficiently to determine
predicts
and density
the wavelength
of the
@1
and would
in
This would provide a useful data polnt in the regime where the "mixing" region
thickness is much less than the be_n radius and whether or not it stays that way with the
instability
OI
run a 115 MA case to t = 100 (3,3 ns) In about 15 CPU hours, but we ex-
current.
22,5 kA-
present.
32
E
O _
REFERENCES
33
O
APPENDIX
Q
LOW-MODERATE
1989.
A.i
BEAM, CURRENT
SIMULATION
RESULTS
AT AFWI
Following
isanexcerptfrom thequarterly
progress
reportforSeptember1988-April
SIMULATION
RESULTS
To reduce computation
l-D"
model to generate equilibria. Cases for plasmoid propagation with ion currents ranging
from 22 kA to 1.8 MA have been run thus far. This corresponds to the number of skin
Q
depths across the beam, wb,r_/c, in the range 2.9 to 26.0, respectively.
of simulating higher currents, we noted that noise levels in the diagnostics increased with
the current to a point that made interpretation
very difficult by 1.8 MA. Increasing
the current further
O
levels.
results.
We, therefore,
and/or
that is being
approximately eliminated)and
in the radial direction (X2). Most of these tests were done
at 22 kA where the run time was short but the noise level still easily observable. The
"current smoothing"
of particles-per-cell
namelist parameters
were adjusted
separately
or concurrently.
parameters
We conclude that
having
This set of
increased the running time by about a factor of two to four over the original
acceptable cost. We then set up a run at a current of 7.2 MA which had run on the AFWL
CRAY to about one-fourth
O
of the equilibration
account was shut down due to lack of funds. Results at that point were not close enough
to equilibrium
(Our expectation
A-1
O
charge
of further
funding of this
the
A.I.1
LOW CURRENT:
Laboratory.
case. 1 Changing
We
the run ID to
the smoothing
NSM2 = 0). The results from run AB were very noisy and considered
unusable
(Figs. A-
lb-A-3b).
Although the results from run AB were not good, it did show us that smoothing
in the radial direction was crucial and had to be used to reduce noise. Run AL was created
from run AA by reducing
smocthJng
2 and turning-off
slice was plotted for the field plots, while five slices were plotted
only two cells for the axial integration.
el
time-averaged
The
behavior.
Because of the fast running time of run AL, it was felt a parameter
done quickly and efficiently using this case. Nine runs were created
different parameters
only one
the axial
study
(AC-AK)
could be
by varying
these runs show little or no change versus run AL at this early time and were not included
in this report.
A.1.2
MODERATE-CURRENT:
For the h = 452 kA runs in this series, we used run AP (previously ran at LANL) as
our base case. 1 Ali runs in this series (BAl, BC-BG, BI) used two axial cells, except for run
BA2, which used four axial cells. Simulation runs in this series were carried out to different
Run BAl was the first run in the series and was used to compare with LANL run
AP. 1 The results of BAl were much noisier than AP (Figs. A-ha-A-10a).
built from BAl by changing IQ = 2 to IQ = 1 (the IQ parameter
count for subcyciing
the particles;
higher numbers
resolve cyclotron
motion better)
increasing the number of axial cells to NX1 = 4. Also the radial smoothing
that
increasing
was created
the number
are quieter
was turned-
it
of axial particles
and
and shows
Run BD
were allowed in the first row of cells along the axis. The results of run BD (Figs. A-hdA-10d)
by reducing the axial and radial cell size from 0.2 cm to 0.15 cm. The results from run
BE (Figs. A-he-A-10e)
are quieter
need for a reduced cell size. Most runs to this point used radial smoothing
but had the axial smoothing
Q
turned-off
(NSM1 = 0).
by setting NSMI=
(NSM2 = 3),
was
reduce noise was to increase the number of axial and radial particles per cell, run BG was
created in this manner.
(Figs. A-hc-A-7c),
reduce noise.
By comparing
it is apparent
that
increasing
the number
of particles
to run BC
per cell does
by
decreasing the number of axial particles per cell by one half. The results showed an increase
in noise and further
Q
of having a sufficiently
A-3
A-7a) shows the importance of having a sufficiently large number of radial particles per
cell. Figure A-11 shows time-average current profiles for runs (BA1-BA2,
O_
BC-BF)which
ran to times > 1600. Again, ali the runs exhibit nearly identical time-averaged
behavior,
A.1.3
HIGH CURRENT:
Laboratory account
OI
per cells radially, and the results looked similar to runs BAl (Figs. A-5a-
results that were of much use was the time average current profile (Fig. A-13) taken at
ct = 2000 cm.
A.I.4
O/
CONCLUSION
0]
!t is apparent that several parameters have a great influence on reducing the amount
of noise in a run. Having the axial and radial smoothing (NSM1 and I_iSM2) turned-on is
crucial to noise reduction, and also IQ (the integer iteration count for subcycling particles)
must be set to a value ,___2. The reduction in cell size was important in improving the
simulation results, Having a sufficiently large number of particles per cell appears to have
had the most noticeable effect of all.
in two
figures. Figure A-14 shows the time average current profile for different values of the beam
current,
current.
Figures A-16a-A-16b
Oi
of
spectrum of the net current for run AA with 129 axial cells with a beam current of 22.5 kA.
Figure A-16b is a time history and frequency spectrum of the net current for run AL with
two axial cells with a beam current of 22.5 kA. After examination one can see that the
A-4
'
number of axial cells has only a slight effect on the noise. Figures A-17 and A-18 show
time history and freqt_ency spectrum
in Figs, A-14-A-15
are used,
as the beam
values used
were taken from the time history plots shown in Figs. A-16-A-18,
Figures A-16--A-.18 also show that the net current reaches a steady state in a time of order
ct/r_ ~ 200c/wb0r_, Thus,
a rough estimate,
tru,(hr)
the number
(NPC/40),
As
to scale as
By taking an overall view of the runs done in h = 452 kA series, run BG (Figs. A5g-A'7g)
At later times
at early times
gave very
A simulation
would take about 4 hours of Cray time to execute out to a time of ct = 400 cm,
A-5
@
2,25
c1_1, 10
/
-0
05
-1
_L.
,0
13,
27,2
4.1 ,4.
55,5
X2
(a)
,
.
.
,,
i....
--1 .0
13,
27,2
41 , 4.
55,5
x2
(b)
_
2.23
_o94
-0,
,,
,,
_.,
,,,,,
36
-I 6_
X2
(c)
Figure
A-I.
Periodic
simulation
results
comparing
Be versus
ru,,,withIb--22.5kA,(a)runAA,(b)runAB,(c)runAL.
@
A-7
;._ _-2
9 .....
I .....
0,8,3
@I
r"
-0,30
"7
_0
26 1
--0,8_
"1 , 0
131,
27
,2
X2
Ca)
I 0.,.
_--"
I..........
4.1 , 4.
........
,
-o,41
.,107--2
1,
6
I
1,3, 1
I
27.2
X2
Cb}
....I
0,89
u-_O
' 02
-0,84
--1 ,71
--0.5
0,57
,
-0,9f__O,o
55
--
I
....
4.1 ,4.
55,5
Ii
I
1
"
--
@-
13,5
27,5
4.1 ,5
55,5
X2
(c)
Figure
A-2,
Periodic
simulation
results
comparing
runs with l b= 22.5 kA. (a) run AA, (b) run AB, (c) run AL,
A-8
,_1Q -_
J 1-IIqT
.... __._
._,_6[1-_--':' .......
--i-_
..........
'
2,23
1 ,20
-0,8_
o,18,.,
,g
i_,5
1,O9--3
.
2
2 .........
I
2_,5
X2
(a)
J 1-1NT1
......
I-":
I y
41,5.......
_,_
'
'
.........
'i
Zl ,02
_0,07
'
I
I
I
-o,8_............
:--,s _3,s X227 s ,_1s -_-ss,_
Ib
Cb)
_31,Q09
'3
--
1 ,99
JI-INTI
i
I'
--
i--
z_o,88 -,
I
...%
-0,23
I
_ / _,_. ._,_
J ,.__;,._
....
, _,_..
--_,__,_-_
x2
Cc)
Figure A-3,
-0
O_
LLNLPLASMOIDPROPAGATION
._
12
"_
RUNAA
01
10 ................................................................................................................
_72_......................
i ...........
---t ....
RUNAB
"
0'-I0
1
20
I
3O
I
4O
[
_
OI
60
RADIUS
(cm)
,A-ll
"
Q
LI. L PLASMOIDPROPAGATIOIq
,
RUN,BAl
40
,_,
RUNBA2
II
<
,5
FZ
,,,
rx
tr"
RUNBC
20 .....................................................................
+
RUNBD
10 .............................................................................................................................
C3
RUNBE
,. ,
RUNBF
-10
' II
-20
0
I
10
I
20
I
39
I
4O
I
5O
60
RADtUS
(cre)
0
Figure A-1I,.
Time averaged current profile in steady state for runs at I_ = 452 kA.
ti
II1
A-17
II
IdOL PLASMOIDPROPAGATION
100
II
8O
Ib
-20
-40
I
1,0
I
20
I
.30
I
40
I
50
60
RADIUS(cre)
Figure A-13. Time averaged current profilefor run BB near end of"run with A = 1808 kA.
A-19
[LNL PLASMOIDPROPAGATION
-40
0
I
10
i
20
I
30
I
40
i
50
II_,
60
RADIUS
(cm)
Figure A-14.
Time averaged current profile comparing different It, for three runs,
A-20
O
I
,,
LLNLPLASMOIDPROPAGATION
0
2O
0
0
1
500
I
lE +03
1
1.5E+0,3
2E+03
Ibecrn(kA)
O
Figure
A'15.
A-21
root of beam
current.
PLASMA
g, _7
SHEATH
RUN AA,
Vl_,,,,e3,
N_,,,O,2,,NP,I1
9
4_
,E9,
VTH(2)-.C)
.30
7,40
* g4
247
--
'o0_0
lo
o,7_
_
_._o
_P
_,2_
L_M
.,o
1o 7
OI
,_3oo
1
]_
@J
':
,o_
,oo_
1 ,g6
_ _l_,.,.
J[ j_,
_.,.,L,_,,
L,A.__k,A,
.t g3
F'REQ
(a)
_ -._
11
5,89
7 8.=,
1O - l
*_
T,Mm-299e,ooooo
@.
U..,K
1_ 1 CellS,
-;.
A-22
II
PLASMA
SHEATN
RUN
44., 4.
BF',
:i ...."
VB-.,,6..3,
NB-.O,2,,,NP-,2,E:IO,
'_ '
I...... -_
""I.....
3,..3,
22,
O0
10 10 _
............
0
50
I
..........
e'4
............
......
1 50
-
'"'
,,10
23
1_
m,
'7
0
0,4.9
O,97
F'REQ
1 ,4.8
TIMEu
II
1 ,95
1999,37000
'@
1 00
lO g
10
O-
VTH(2),,.O,30
,_. 32,0
A.23
OI
PLASMA
SHEATH
RUN
BB
19g,
VB',e3,
XlB"_
_
4 P2,NPuS,EtO,
VTH(2)'0,30
II
j
138,
76,
15,
-4(_",00
'
0 50
'
10 11 _'_--:'.....- .... I
1 ,00
;]::,LMI_
........._Pr__1,r_lJM
.......
1 ,50
,10
2300
.....
-....
_......
I...............
'_
10 10
10
1o 6
5
100,00
i_
0.98
1 ,g6
FREQ
2,95
TIME-
3,9.3
1999,2OO00
lt
Figure A.-18, Periodic eimuliLtion BB_ time history aild frequency spectrum of the net
current, Ib = 1808 kA, Current in units of 1,35 kA, frequency in unita of
wo = 3 10I rad/s,
lJ
IL